[abc] response digest

Hi, I've been editing some of the responses for general consumption. More
to follow... don't read the parts that I forgot to edit away...

-O

Gregor Kiczales: (British Columbia)
---------------
I will forward it to the lab, I bet some of the students will
want to try it.

Right now, all my AspectJ development is on small programs in Eclipse,
because I'm preparing the two-day AOP course, and Eclipse integration
is essential for that.

Gregor

Mary Lou Soffa: (Virginia)
---------------

Hi Oege,

<... snip ...>

Good news about you AspectJ compiler. I am interested
and someone here - Kevin Sullivan - would also be
interested. I will tell him about it.

<... snip ...>

Best regards,

-mary lou

Don Batory: (Texas at Austin)
-----------
Howdy!

Living in Texas would be better if W retired here permanently real soon.
Other than that, good.

I'll have a look.

In the meantime, please look at this paper (which my student just
completed).
It summarizes exactly the kinds of additional things that you'd want
in your tool, and gives reasons why. It's not directed to the AspectJ
crowd, but the ultimate usability of some of its capabilities.

don

Chris Dutchyn: (British Columbia)
-------------
Gregor passed along your note giving the location of abc; it is a tool
that you have mentioned to me before.

In briefly glancing over the site, I found a typo in the Differences page,
sixth bullet point: there must be a missing right parenthesis in the
second of the two distinguished pointcuts.

--
Christopher Dutchyn
UBC Computer Science
Mario Sudholt: (Nantes)
--------------
Dear Oege,
as suggested I will pass your request to my colleagues some of which
have done some non-trivial applications with AspectJ. I will also give
it a try (even if I have not done much AspectJ programming myself).
All the best,
    Mario.
-- 
Mario Südholt
Researcher, OBASCO group, École des Mines de Nantes/INRIA
www.sudholt.net
Todd Millstein (San Diego)
--------------
Hi Oege,
Congratulations on getting your compiler almost finished!  I'll definitely
take a look.  Unfortunately, I don't program much in AspectJ, so I don't
have a bunch of my own AspectJ programs readily available.  So I don't
think I'll be able to stress-test your compiler much.  That being said,
I'll certainly keep abc in mind as a platform for my future research.
About Relaxed MultiJava, I haven't thought particularly about how to
integrate some of its ideas into AspectJ.  However, I'd be very interested
to hear what you are thinking about and to discuss ideas with you.
Hope you're doing well.
Todd
Jianjun Zhao (Fukuoka)
------------
Dear Oege,
Thank you very much for your great information!
I will download the ABC implementation and let my students
to try it. We will let you know our findings on ABC.
I am still interested in slicing AOP, and my student is now developing
a slicer for AspectJ using Soot framework based on bytecode (Jimple code)
analysis. The free available of ABC is definitely very helpful for our
recent implementation.
Looking forward to the public release of ABC!
Best regards,
Jianjun
Shmuel Katz: (Technion)
------------
Dear Oege,
Thanks for sending me this information on abc. I am trvelling most of this month
and it is still vacation here for universities. But I will discuss this with
some students, and try out your compiler/tool. (However, it might be a while
before we have feedback.)
I also have a student working on static identification of types of aspects....
I leave today for 9 days in Japan, as the first of my September trips, but hope
we can stay in touch.
Yours,
 Shmuel
Paul Kelly: (Imperial)
-----------
Hi Oege
Good to hear from you - and it's great to hear your compiler project is
making good progress.
I'm personally tied up with some other things over the next few weeks.
I've forwarded your message to a couple of my people - though we haven't
been doing very interesting things with AspectJ I'm afraid.
Paul K
Hidehiko Masuhara: (Tokyo)
-----------------
Dear Oege,
Thanks for the information on abc.  I'm very happy to try it out and
to provide feedback if any.  Actually, my students and myself have
several plans to implemet extended AspectJ languages or compilers, so
that will be very strong platform candidate!
Best regards,
Hidehiko
Harold Ossher (IBM Watson)
-------------
Dear Oege,
This sounds like fantastic progress with abc! I hope you have success with
the tests on AspectJ programs (I can't offer to help there).
Things are going well with the CME, thanks. A beta version did go out on
Eclipse open source a few weeks after AOSD, and we are working on
enhancements and on writing. I've attached a paper we just finished on the
composition concepts in the CME. They're the next generation of those used
in Hyper/J, considerably less ad-hoc. When you get to looking at support
for our approach, I suggest you look at this before (and perhaps instead
of) Hyper/J itself. The paper describes the concepts and also gives a brief
sense of how the CME Concern Composition Component implements them, with a
number of open points. It would be interesting to chat at some point about
how the open points provided by abc and CCC compare.
I should mention that CCC is a component in the CME, not an end-user tool
like Hyper/J. Instead of exposing a composition language for developers, it
exposes a composition interface for use by tools. This interface has a
textual form (Plainway, described a bit in the paper and in detail in
documents on the cme web site), but that is lower-level than we would
expect most developers to want to use directly. We're currently working on
higher-level specification of composition for direct use by developers.
However, for your purposes, the CCC/Plainway level is probably of greater
interest.
Regards, Harold
(See attached file: Concepts for Describing Composition of Software
Artifacts - distribute.pdf)
Curtis Clifton: (Iowa State)
---------------
Oege,
On Sep 3, 2004, at 10:25 AM, Oege de Moor wrote:
> [snip]
>
> At this stage it would be tremendously helpful to us if a few
> professional friends could try abc on their own AspectJ programs,
> and report potential problems on our bugzilla database. A
> pre-release of the compiler is available at
>
> http://abc.comlab.ox.ac.uk
>
> You will also find related links on that website, for instance to
> subscribe to abc-related mailing lists.
I'm very interested in this work and I've just subscribed to the abc
mailing lists.  I don't have much of an AspectJ code base on which to
test your compiler.  Most of my aspect-oriented code consists of small
case studies with additional annotations that are processed by my own
AspectJ extension.  My extension uses ajc as the backend, so I could
probably plug in abc instead.  Though that will have to wait a few
weeks (at least until after that Sept. 30 deadline that I suspect you
are also eyeing :-).
> [snip]
> Now that we have a solid AspectJ compiler, one of the first
> steps we want to take is to integrate the ideas of MultiJava
> (in lieu of the baroque intertype decls of standard AspectJ).
> Is this something you've thought about? I'm keen to explore
> possible collaboration in this area...
The integration of MultiJava's open classes and multiple dispatch
design with AspectJ's advice mechanism is something that I am very
interested in.  Unfortunately it is outside the scope of my
dissertation and so I've set aside the MJ/AJ integration work for the
time being.  I too would be very interested in collaborating on this
work, but will likely not be able to devote much time to such a
collaboration until late spring.  In the interim, I would be able to
participate in email discussions, but not in any serious coding or
experimental efforts.
Cheers,
Curt
Kevin Sullivan (Virginia)
--------------
Hi Oege,
You might remember that we met recently. Mary Lou forwarded me your e-mail,
below. Congratulations on the release of your new compiler. We're planning
on having a look at it. You might be interested in AO language design and
implementation work work that my PhD student, H. Rajan ("Rajan"), and I have
done recently. In a nutshell, we're aiming for greater conceptual integrity
and compositionality in AspectJ-like AO language design. We've got a
language design and a working compiler for it--it extends C#. I'm attaching
a draft paper. Hope you're well!
Kevin
Shigeru Chiba (Tokyo Tech)
-------------
Hello,
From: Oege de Moor <Oege.de.Moor@comlab.ox.ac.uk>
Subject: abc
Date: Fri, 3 Sep 2004 16:08:01 +0100 (BST)
> At this stage it would be tremendously helpful to us if a few
> professional friends could try abc on their own AspectJ programs,
> and report potential problems on our bugzilla database. A
> pre-release of the compiler is available at
>
> http://abc.comlab.ox.ac.uk
>
> You will also find related links on that website, for instance to
> subscribe to abc-related mailing lists.
Wow, it's a great job!  Unfortunately, my group aims AOP-based tool
development and thus we do not have many AspectJ programs for testing
abc.  However, I'll circulate the above URL in my group; I think
some of my students should be intrested in abc.
Thank you and best regards,
Chiba
Paul Kelly (fwd)
----------
Hi Susan, Hi Naranker,
I mentioned Oege de Moor's AspectBench compiler over lunch today.
Here's the email - it's semi-private at present.  I don't have anyone
with time to play with this so if you do have a look please drop Oege a
line.  We should get Oege to come and visit...
Paul K